Subspace Pursuit for Sparse Signal Reconstruction in Wireless Sensor Networks

Subspace pursuit (SP) algorithm provides the two most important features of greedy algorithms. It has lowest computational complexity in comparison to other greedy algorithms such as OMP and ROMP. It provides the reconstruction quality of same order as that of linear programming techniques for very sparse signals. SP algorithm estimates the sparse signal step by step, in an iterative fashion. The presented analysis shows that SP exactly recovers the sparse signals for temporally correlated data collected during real-time experimentation using NI WSN platform. The efficacy of recovered signal is analyzed using the parameters of peak signal to noise ratio and root mean square error.